Robotics recently spread to spine biomechanical research. The aim of the present work is to describe and validate a new method for in vitro studying of a multisegmental spinal specimen under dynamic conditions. This method relies on the use of a simulator with six degrees of freedom (to impose movements in all directions), an optoelectric apparatus (for collecting kinematics data) and an original system for attaching kinematic markers, allowing their precise removal and replacement under different examination conditions. The accuracy of measurements as well as their reproducibility under static and dynamic conditions is reported here in the study of a human lumbar spinal specimen (L1-sacrum). The method appears to be reliable and reproducible, and should therefore enable future studies of variations in mobility between healthy and pathological spines, to better understand the influence of different implants on spinal kinematics.